Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
Munich Personal RePEc Archive
Econometric Models used for the
Corruption Analysis
Andrei, Tudorel and Stancu, Stelian and Nedelcu, Monica
and Matei, Ani
Academy of Economic Studies Bucharest, Academy of EconomicStudies Bucharest, Academy of Economic Studies Bucharest,National School of Political Studies and Public Administration(NSPSPA)
15 January 2009
Online at https://mpra.ub.uni-muenchen.de/19623/
MPRA Paper No. 19623, posted 30 Dec 2009 10:03 UTC
����������������� ����������
��������������������� ��������
�������������������������� ������������
����������������� !��������� ���������
�������������������������
��������������������������������������"����� ������������
�
�
��# #��������#����������$#���%���#������# �
� ��&�����
� �� ������������� ������ �� ���� �� ������� ��� � ��������� ��� ����
������������������������������������������������������
������ ��� ��������� ���� ��������������� ����� �� ������� ��� �����
����������� ������� �������� �� ��������� � �� ��� �����������
�������������������� ���� ����������� ��������������� �������
����������� ���� ������������ ��������� ������� ����
����������� ���������������������������������������������
���� ��� ���� ������� ������� ���������� ���� ���������������
������������� ���� ������� ��� ���� � �� ��� ������ �����������
��������������
� '�!� (���) ����������� ����� ���������������� �����������
�������
�*�����������������)��+,��%-.�
/0� � ��#�����# �The corruption analysis is a significant field of the economic research
nowadays. Over the last years, the university and reseach environment, the
international organizations (such as the International Monetary Fund, the World
Bank and so on) have showed an increasing interest in estimating the corruption
level, in identifying the causes, the mechanisms for transmiting it among the system
and also an interest in measuring its impact over the economic and social processes
of a country or development region.
Among the most important classic writings in this research area there are
A. Krueger [1974], S. Rose-Ackerman [1975], Mauro [1995], Tanzi [1998] and so
on. The period of time following the political and economic changes caused by the
socialism’s collapse in the Eastern Europe caused the appearance of new
corruption forming factors and more and more refined mechanisms of transmiting it
among the new transforming social systems. These new conditions stimulated the
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
research development in this field. Furthermore, the World Bank (2001) identified
the corruption as „the only significant obstacle in the economic and social
development”. Over this period of time there have been published several
remarkable articles concerning the quantitative analysis of the corruption
economics. Some of the most important writings from the 90s are written by P.
Mauro [1995], P. Bardhan [1997],�V. Tanzi [1998], Shang-Jin Wei (1997).
A major aspect in analysing the corruption is identifying its causes in a
public system. In the specialized literature there are identified four groups of
factors that influence directly the corruption among a system: political, legal,
historical, social, cultural and economic ones. The quality of the political system,
the features of the legal system (Leite and Weidmann (1999)), especially the
legislation and the institutions concerned with abolishing the corruption, the quality
of the democratic system, the features of the electoral and the administrative
system in a country fall into the category of political and legal factors.
A series of studies such as La Porta (1999), Treisman (2000) emphasize the
influence of the traditions and historical factors over the corruption level in a
country and the features of the mechanisms of forming and transmiting it. The
social and cultural factors are significant in emphasising the corruption features of
a country (La Porta (1999), Treisman (2000), Alesina (2003)). On the other hand,
the religious factors also have a significant part in spreading the corruption
throughout a social system. The economic factors, such as the opening level of an
economy (for instance Dreher (2003), Treisman (2000), Wei (2001)), the
dimensions of the public sector (Tanzi (1998), Treisman (2000)), the salary level in
the public sector (van Rijckeghem (1997)) etc. have a direct influence over the
corruption level existing in a country.
Another major aspect of the corruption analysis is chooosing the most
appropriate econometric models in order to estimate its effects over some activity
sectors. Among the most important research directions which are concerned with
estimating the consequences of the corruption on the social and economic
environment there are: (i) directions regarding the economic growth in general
(Mauro [1995]��Abed and Davoodi [2000], Krueger [1974]); (ii) directions with
reference to some national economy sectors (Tanzi [1998], Shang-Jin Wei [2001]);
(iii) directions concerning the effects of the decentralization process on the level
and the mechanisms of transmiting the corruption throughout a system (Shah
[2006]) and so on and so forth; (iv) directions regarding the quality of the
finance systems of some activity sectors such as the military one, Gupta [2001], the
salaries in the public sectors (van Rijckeghem and Weder [1997]); (v) directions
concerning the industrial policies of a country (Emerson [2002], Bhagwati [1982])
and the efficiency of the investments (Sarkar [2001], Mauro [2002]). In most of the
above mentioned writings, in order to estimate the impact of the corruption on
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
some economic and social aspects, there can be used regression models, VAR
models, the analysis of integration series etc. Kaufman [1999] and Andrei [2008]
propose using the simultaneous equation models with the view of estimating the
effects of the corruption on the public administration processes in a country. In this
case, the variables of the model fall into endogenous and exogenous ones, and the
parameters are estimated by using the method of least squares in two stages (TSLS)
and the general method of moments (GMM). By using these methods, the
endogenous or exogenous type of each variable from the model is taken into
consideration. With reference to this model we must mention the fact that in the
specialized literature there is no convenient approach of choosing the tool variable
list used to estimate the parameters of the simultaneous equations models utilised
for the corruption analysis and its effects on the public administration reform
(Andrei [2007], Profiroiu[2005], Teodorescu [2007a]), the corruption existing in
the Romanian universities (Teodorescu [2006, 2007b]), and the consequences that
the corruption has on the finance quality of some activity sectors (Andrei [2002] Hi
Matei [2007]).
At present, the public administration in Romania is subject to an intense
reform. According to the demands of the UE accession process, the public
administration reform is defined through a series of reform measures which are
being taken in the public sector through carrying on the decentralization process
and improving the process of forming the public policies.
According to the reform strategy of the public administration, the
decentralization process has a significant part in the fight against corruption.
Nevertheless, when applying this process we must bear in mind the fact that the
deficient implementation may lead to a local growth in corruption, with dire
consequences on the social and economic environments, on both middle and long
term.
+0� �%����$� ���# �#$��%���#����In this writing, in order to define the analysis model of the reform process
quality in the public administration, we have as a starting point Becker’s [1968]
approach to crime and punishment. Thus, the choice that one makes to commit a
crime or an offence is inspired by the ratio between the benefits from the action and
the losses caused by not doing it or the penalties suffered by its initiator if detected.
� When defining the analysis model of the reform process quality, you must
take into consideration the quantitative measurements of the results of some reform
actions. Therefore, in this paper the public administration reform is defined
corresponding to the reform strategy of the public administration adopted by
Romania in 2004. We must mention that the administration reform process was
recommended by both internal necesities and Romania’s accession to the European
Union. Thus, writing this document was possible with the direct support of the EU
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
Comittee, by providing Romania through Phare 2001 with financial and technical
assistance by means of the project “Support for realizing a strategy project for the
public administration reform”.
According to the strategy adopted in 2004, the administration reform is a
process of transforming the central and local public administration in order to
comply with the demands of the beneficiaries of the sector and the severity of the
accession to the European structures. From this point of view there have been
identified three constituent parts of the reform process: (i) the reform of the public
function, by improving its management, and the change for the better of the
continuous forming process of the civil servants; (ii) the reform of the local public
administration, by carrying on the decentralization process; (iii) the improvement
of the expressing process of the public policies.
Under these circumstances, we define the public administration reform
process starting from the next assumptions:
1%!��������� ��0� /2� � In order to support the reform process we have a reform
strategy and a battle plan. By apllying the battle plan we aim at achieving three
specific objectives: O1 – on the short run, the actions regarding the public function
have as a goal to improve the management of changes as far as the public function
is concerned; on the long run, the aim is to stabilize and strengthen the public
function system; O2 - in the field of local public administration reform we aim at
approaching the public administration and the citizen and creating new mechanisms
through which the central Government should have a better communication with
the local public administrations; O3 - actions that should improve the quality of
the expressing process of the national and local public policies which would
strengthen the managerial capacity of the Government, the Local and District
Councils, capacity which is useful for achieving the objectives and demands of the
local and national development.
1%!�����������0�+2��With the view of supporting the specific objectives, there are
some precise activities within the framework of the battle plan which comes
together with the reform strategy. In order to measure the positive effects of the AP
reform process we define the following three functions:
1( ,... ), 1,2,3�� � � �� ! ! �= =
where ��! , with 1,..., �� �= , represents the ensemble of planified actions with the
view of supporting the specific objective "�� � . �� means the number of planified
actions in order to support the specific objective.
There are 17 actions for the public function reform which aim at: (i)
creating and applying a recruiting, valuating and promoting system mainly based
on merits as far as the public function is concerned; (ii) creating and applying a
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
unitary salary system for the civil servants; (iii) enhancing the number of
professional public managers and their skills which could serve supporting the
reform process and the accesion to the EU; (iv) consolidating the capacities of the
National Administration Institute with the view of ensuring the application of the
strategic parts of the continuous forming of the public servants. Besides that, there
have been provided 25 activities for the field of local public administration reform,
activities which concern: (i) the definition of the mechanisms and structures that
are necessary for managing the evolution of the process; (ii) the reform of the
Education, Health and Social assistance sector; (iii) the improvement of the local
budget system; (iv) the income enhancement and the elucidation of the issues
regarding the public property; (v) the improvement of the transfer system; (vi) the
establishment of a new frame for the Prefect’s activity and the development of
some specialized training programmes for them; (vii) the development of the local
authority capacity of undergoing the reforms by training the human resources,
developing the standards and administrating the new decentralization servicies. In
order to support the improvement of the public policies formulating we have
provided 11 actions which aim at: (i) consolidating the government capacity on
both central and local levels with the view of supporting the public policies
formulating process; (ii) enhancing the role of the most important public servants
in the public policies formulating process; (iii) improving the public policies
formulating process; (iv) strengthening the coordonation between the institution
both on central and local levels in the public policies formulating process;
1%!�����������0�.2��On middle and long term the reform actions that are part of the
strategy will bring a benefit to the public administration by improving the quality
of the services offered to the beneficiaries, reducing the costs of the public
institutions functioning, supporting the development process on a national and local
level, diminishing the corruption etc. Under these circumstances, the result will be
that each function which values the positive effects of the reforms on the three
fields will be directly influenced by the reform dimensions measured through the
activity volume destinated for each action within the strategy:
0�
��
!
∂>
∂ 1,2,3 Hi 1,2,..., .�� � �= =
1%!��������� ��0� 32� �With the aim of supporting the reform process, financial
resources were provided for the reform actions. These came from the state budget
or external projects, mainly European ones, part of the Phare Programme which
lasted from 2004 to 2006, or the PAL Programee of the World Bank. The financial
resources fall into three priorities: (i) the public function: 7.200 thousand Euros.
Besides that, 13.937,6 thousand Euros per year are added in order to form the
following personnel categories over a period of three years: 100 major civil
servants through specialization programmes lasting one year; 150 young
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
professionals through specialization programmes lasting one or two years; 3000
actual top management civil servants through specialization programmes lasting
one year and 9000 actual middle management civil servants through specialization
programmes lasting three months. (ii) supporting the local public administration
reform: projects with Phare or the World Bank finances of 8.150 thousand Euros;
(iii) the third part: projects totalizing 8.150 thousand Euros. Moreover, from 1992
until the start of the 2004-2006 Phare programme the financial support from the
European Union for the public administration reaches the sum of 42.000 thousand
Euros. The value of the Phare projects from the 2004-2006 Phare programme
which support the Strategy totalize 35.880 thousand Euros.
As far as the model used for analysing the reform process is concerned, the
reform costs are given by the activities developed in order to support the reform
process, by the positive reaction of the public administration system to the reform
measures provided by the strategy document and by the system’s losses due to the
lack of political intention to apply some reform measures. Under these
circumstances we define the cost of the reform process for each priority through the
functions:
1( ,..., , , ) 1,2,3�� � � ��# # ! ! �= =� �
The vectorial variable 1( ,..., )�$ $=� values the positive reaction of the
system to the reform process and the vectorial variable 1( ,..., )�% %=� describes the
political system’s capacity of supporting the public administration reform process.
The vectorial variable � depicts the influence of various factors such as: the
mobility of the public administration employees, the ability of the public
administration to properly organize the competitions for filling in or promoting in
the public function, the level of the informatic systems equipments within the
public administration institutions and so on. The vectorial variable � describes the
direct or hidden actions of the political system over the public administration, such
as: political insecurity in promoting some reform measures, the pressure of the
political system on the public administration through different decision channels in
order to gain some advantages, the corruption induced in the public administration
by the central and local political class etc.
Under the circumstances of a normal reaction of the public administration
system to the reform process, we are going to consider the functions
( ) 1,2,3� �# # �= ⋅ = , which have the following properties:
4�/5�The larger the reform process, the bigger the costs of the reform process,
represented by the amounts of money provided by projects either with national or
international finances:
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
( ) 0.���
�
## !
!
∂′ = >
∂
4�+5� �The reform measures given by the strategy document ensures to a certain
degree positive changes in the system. Thus, there is a series of risks which can
slow down the reform process of the public administration. On the other hand,
there is also a series of positive reactions, unpredictable when elaborating the
strategy, which could diminish the reform costs and obtaining positive results
within a shorter period of time than planned. Under these circumstances, the
function used for measuring the costs will be decreasing and concave:
2
2( ) 0 ( ) 0.� �
�� ��
� �
# ## $ # $
$ $
∂ ∂′ ′′= < = >
∂ ∂
4�.5��For the vectorial variable �, which measures the political will of the political
class to support the AP reform process, we will have one of the following
situations:
( ) 0���
�
## %
%
∂′ = >
∂
if there is no political will to reform the AP, in which case the costs of the AP
functioning are much higher than the situation of reforming it or
( ) 0,���
�
## %
%
∂′ = <
∂
if there is political will among the political class and this one efectively and
efficienly communicates with the technical level from the APC and APL in order to
apply the reform measures.
1%!�����������0�62�The net income obtained by developing the reform process is
defined by the function:
1 1( ( ,..., ) ( ,..., , , ))�� � �� � � ��
�
#& ! ! # ! != −∑ � �
Actually, the net gain resulted due to the application of the reform
measures is found in:
4�/50� The increase in the AP capacity of accomplishing its stipulated basic
functions.
4�+50�The diminish in the AP and national corruption level.
4�.50�The increase in the AP and national transparence.
4�350�The increase in the employees satisfaction within this activity sector.
Each one of the four endogenous variables are established by a series of
factors. Within the study, taking into consideration the data series recorded from
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
the AP employees, there will be estimated the regression models parameters used in
order to analyse the four endogenous variables. With the view of defining the
models and apllying some appropriate methods so as to estimate the parameters, we
must bear in mind Kufmann’s [2002] study. Thus, according to it, in the majority
of the studies carried out in the public sector there are two major limits, which
concer the models used for analysing some fenomena from this activity sector and
the methods used to estimate the parameters. More exactly, the first objection
refers to the fact that most of the studies which were carried out are based on
regression models estimated on cross-country data. Secondly, the methods used to
estimate the parameters which do not take into account whether the variables are
endogenous or exogenous are severely criticized.
.0� �%�������������������7%��%����������In order to determine some features of the public administration reform
process in May 2007, a research based on a statistical poll among the public
administration was carried out. The research used a representative sample from the
civil servants working in the public administration. In order to create the sample we
used a two-stage method, and its volume reached the number of 971 civil servants
from the central and local public administration. The error of the parameter
estimation related to the reference population is of 1,2 percent and the probability
that the results are truthful is 97 percent.
A statistical questionnaire was applied to the observed population,
including questions which fell into the following major themes: the internal
organization of the public administration institutions, the decentralization process
in the administration, the public function, the discrimination within this activity
sector institutions. In the questionnary there was also inserted a series of questions
regarding personal aspects, such as one’s gender and age, preparation level, the
kind of institution in which they activate and so on.
Starting from the questionnary structure, there were defined three
cathegories for the variables based on their agregation degree.
1. The first level variables have the lowest agregation degree. Each variable
from this cathegory describes a certain degree of the reform process. The values of
these variables are determined directly on the basis of the data recorded from the
sample. The notation used for them in this writing is the letter ....'
For instance, 5.22' values the degree to which the corruption existing in the system
benefits from the shortcomings of the salary system of this activity sector. Besides
this variable, on this agregation level, there is also a series of variables which refer
to: the legal framework shortcomings which encourage the corruption within the
system, the civil servants liability to do corruption acts, the quality of the economic
and political environment to to ease the corruption acts and the degree to which the
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
citizen behaviour encourages the corruption acts. Except for the variables which
express personal features, the variables from this agregation level are defined on
the basis of four or five values. For instance, the variable used above can take
values in {1,2,3,4,5}. These variables do not explicitly appear in the econometric
models but through the variables which are in the second or third agregation level.
In order to define the questions from the questionnary and at the same time the first
level variables, we must take into account the following: (i) similar studies from
the specialized literature (Kaufman [2002], Profiroiu [2005] etc); (ii) the analysis
of the diagnosis and the reform measures proposed by the public administration
reform strategy for the period of 2004 to 2006; (iii) the content of the Phare
programme which lasted from 2004 until 2006 and supported the application of the
public administration reform process; (iv) the PAL programme of the World Bank
etc.
2. The second level variables are calculated by summing up the first level
variables and define a certain aspect, not very general if related to the first
agregation level variables of the reform process. These are noted with � ( in the
model.
3. The third level variables measure a general aspect which concerns the
public administration functioning and the implications of the reform process over
it. Variables on this level have the highest agregation level and result from the first
and second level variables. These are noted with # in the model.
The second and third level variables are defined in a way which allows
creating a simultaneous equations model that may analyse the following four
aspects of the AP reform process at the same time: the AP capacity of
accomplishing its basic functions stipulated by the law system, the analysis of some
aspects dealing with the transparence of the central (APC) and local (APL) public
administration institutions and the examination of the influence factors over the
satisfaction of the employees activating in this activity sector.
The most important variables of the research plan are presented in the
Appendage no. 1. For each variable we are going to provide a synthetic definition,
as well as the number of the first and/or second level variables that were used for
calculating it.
30� �%����8�����# ��#������
30/0�����!���� ������������������������������ ��������� ������������
Two regression models are defined using the variables from Annex I to
analyse the institutions performances from AP. In both situations the endogenous
variable is the one which quantifies the quality of the activity carried on in the
public institutions from AP. The parameter estimation, considering the data from
each institution, realised in four different situations: (i) public administration level
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
(AP); (ii) prefect’s offices (P); (iii) district councils (CJ); (iv) decentralised offices
(SD).
The first regression model analysed the public institution performances on
four levels, considering the corruption level and the quality of the courses
organised in the AP institutions to promote in public function:
1 1 5 5 9 9 1# � � # � # �= + + + [M1]
To estimate the parameter it was used OLS method and was considered
only the records (questionnaires) with all the responds. From the total of 971
persons, only 813 mentioned the institution where they work. Using these records
had been estimated the model parameters at AP level. If it was took in account the
records without the institution name, the total panel will have 873 records, but the
estimation are not significant different. The results are presented in Table 1,
column 2, values being marked with T.
The second regression model has as explanatory variables: one which
measures the corruption level, others which characterise the decision transparency
at the level of the institution from AP, satisfaction degree of the employees and the
influence of the intern and extern factors which can reduce the corruption:
1 1 2 2 3 3 4 4 5 5 6 6 7 7 2# � � # � # � # � # � ( � # �= + + + + + + + [M2]
In the two models, the variables 1� and 2� � are residual with mean zero and
constant variance. These quantify the influence of other factors not included in the
model from objective reasons – for example the performance of the institutions
from central and local administration.
The parameter estimation was realised in the same conditions as for the first
model. For this reason the initial number of values has been reduced from 953 to
878. The results are presented in Table 2.
30+0�����!���� �������������������������� �����
We will define a regression model to analyse the corruption level 5( )# �in
comparison with different factors of influence (i) factors which contribute to the
increase of the corruption level. In this category we include: competitions` fraud
from FP 9( )# � and the pressure of the politic system. 6( ).( (ii) factors which
contribute to the reduction of the corruption level. In this category we include: the
quality of the activities developed in the AP 1( ),# � institutions, transparency from
AP 2( ),# satisfaction degree of the employees from AP 3( ),# quality of the
relations between workers, actual capacity of AP 4( ),# to carry out its functions
and actual 8( ),( capacity of the system to finance the public services 3.21( )' �and
the quality of the reform process of FP 10( ).# . (iii) the characteristics of the
workers from AP. We include three variables in the models: gender 11( ),# ,
management or execution category 12( )# �and educational background 13( ).# .
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
The regression model is defines as follows:
5 0 1 1 2 2 3 3
4 4 5 3.21 6 6 7 8 8 9 9 10 10 11 11 13 12 14 2
# � �# � # � #
� # � ' � ( � ( � # � # � # � # � # �
= + + +
+ + + + + + + + + +
[M3]
where 2� is a residual variable which quantify the influence over the corruption
level of the factors not included in the model.
Parameter estimation was realised at the level of AP, PAC, prefect`s offices, CJ
and decentralised offices. For each institution was considered only the
questionnaires with valid responses to all the variables included in the model. The
method used in estimation was OLS and the results are presented in Table 3.
30.0�����!���� �����������������������!������������������������
To define the model used for analysing the transparency can be choose one,
two or more variables 2# ���� 2.#′ as endogenous. The explanatory variables used in
the model can be classified as follows: (i) corruption which encourage the
transparency reduction into the public institutions; (ii) variables which describe the
specific behaviour of the employees from AP, including here: satisfaction degree of
the employees, quality of the relations between workers and the correctness of the
promoting process. All these three variables quantify factors in direct correlation
with transparency. (iii) variables which quantify elements of the reform process in
public administration. These are quantifying factors with positive impact over the
increase of transparency in public institutions, if the reform process is felt in the
system and a negative impact, if the effects are negative or under employees’
expectations. (iv) personal characteristics such as: gender, position in institution,
education, etc, which differentiate the persons considering the transparency
perception in the public institutions. In these conditions we define the next
regression model to analyse the transparency in the AP institutions:
2 0 1 1 2 5 3 3 4 4 5 8 6 10 7 12 3# # # # # ( # # �= + + + + + + + + [M4]
The parameter estimation was realised using the OLS method and the results are
presented in Table 4. For each institution was considered only the questionnaires
with valid responses to all the variables included in the model.
3030�����!���� ������������ ���!���9�����������������
The satisfaction degree of the public officials is a result of the conditions
offered at office, the salary obtained as a result of their activities and
responsibilities, the respect gained between the colleagues and the results of the
reform process from public administration. In the same way, the satisfaction
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
perception can be different on gender, management and execution categories, on
each ones education. In these conditions the regression model used to analyse the
employees satisfaction is defined as follows:
3 0 1 2 2 4 3 9 4 10 5 11 6 12 7 13 4# � � # � # � # � # � # � # � # �= + + + + + + + + [M5]
The parameter estimation in the four situations (AP, P, CJ and SD) is
realised using the three methods known, also for the three regression models
estimated above, and the results are presented in Table 5.
60� �#�����7��%�������� �#����:����# ���#�� ��&���
�%���%� #�� ��$�#����
Considering the definition of each variable, the data series available, descriptive
analysis realised above, the regression models estimated and the literature’s
approaches Kaufmann [2002], Bai and Wei [2000], Kaufmann, Kraay and Zoido-
Lobaton [1999] etc, it can be defined a model with simultaneous equations to
analyse the following aspect of the institutions from AP: performances, corruption
level, transparency and the satisfaction degree of the employees. To define the four
equations of the model we will consider:
Institutions performances = f1(.)
Corruption = f2(.)
Transparency =f3(.)
Satisfaction degree of the employees =f4(.)
The model with simultaneous equation can be written in the next form:
1 2 3 4 5 6 7 1(1) (2) (3) (4) (5) (6) (7)# # # # # ( # ε= + + + + + + + [M6]
5 1 2 3 4 3.21 6 8
9 10 11 2
(8) (9) (10) (11) (12) (13) (14) (15)
(16) (17) (18)
# # # # # ' ( (
# # # ε
= + + + + + + +
+ + + +[M7]
2 1 5 3 4 8 6
8 10 10 11 12 3
(19) (20) (21) (22) (23) (24) (25)
(26) (27) (28) (29) (30)
# # # # # ( (
( # # # # ε
= + + + + + +
+ + + + + + [M8]
3 2 4
9 10 11 12 13 4
(31) (32) (33)
(34) (35) (36) (37) (38)
# # #
# # # # # ε
= + +
+ + + + + + [M9]
The endogenous variables of the model are the ones used to quantify the
transparency, system transformation as a result of the political changes, employees’
satisfaction and the influence of various institutions in reducing the corruption
level. The other variables are considered exogenous.
;0� �#��� ��
The utilization of the model with simultaneous equations to analyse the
corruption has some advantages concerning the parameter estimation. In this model
is taken in account the exogenous character of the variables and the parameters
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
estimated are not displaced and consistent. The following aspects are important in
this case: (i) variables separation in exogenous and endogenous; (ii) the right
selection of the estimation method (iii) if for estimation is used least-squared
method in two stages, it is important to choose correctly the instrumental variables
(see Kaufmann (1999)).
The model with simultaneous equation defined above can be identifying because
the number of the excluded variables is higher than the one of endogenous
variables. The model parameters are estimated through the least-squared method in
two stages (TSLS) and General Method of Moments (GMM). Using these analysis
instruments we take in account the endogenous or exogenous character of each
variable. In the economic literature isn’t a suitable approach in choosing the
instrumental variables for the models with simultaneous equation (Bai and Wei
[2000]; Kaufmann, Kraay and Zoido-Lobaton [1999]). In the corruption analysis
using the econometric models and where the parameters are estimated using data
from a simple sample are important the following aspects: (i) correct definition of
the questions from the statistical instrument applied to the statistical population;
(ii) correct aggregation of the primary variables in the process to obtain the
aggregate variables of different orders. In this analysis were used primary and
aggregate variables by order one and two. In Annexe I are presented all the
variables used in the models.
REFERENCES�[1]�"���80����<�����%0�4+,,,5������� ������������������������������ ��
����������� ����������� � ������ ����IMF, Working Paper No 00/132.
[2]��������0�4+,,+5������������ ������� �� ������������� � �������� ���
���� ���� ��� !��������NATO (Brussels), Working Paper.
[3]��������0�������������0�4+,,35������� �"����#���������� �����!���������
��������� ���� ��������������$�����%���������$��������������� ������� ���
The International Journal of Economic Development, volume VI.
[4]���������=>����7�����>*0��4+,,,5, ����&��� �$�������'���������$�����
�� ������������� � ����World Bank Working Paper No. 2575.
[5]��������0�4/?-;5���� ���(����� ������������������� �0�Englewood
Cliffs, NJ: Prentice-Hall.
[6]�� ���������0�4+,,;5������ ���������� ��%������������������� ���
�) �#������* ��� ��� Palgrave Macmillan, London.
[7]� ��������0���0�4+,,+5������� ������ ������ ������� ��� ���������!������
����� ����The Journal of International Trade & Economic Development, p. 63-76.
[8]$��@"������0�4/??A5��������������+�������� �$%�+������������� ���
World Development, No. 25, p. 1029-1043.
[9]8������B���0�4/?-35��������� ������!�� ������ ��� ��� !����)������;
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
[10]8������0����<�����%0�������=������0�4+,,,5������� ������������! � ��
������������������������ �������! �����IMF Working Paper, No. 00/116.
[11]8������0����@�������������������0�4+,,/5������� ������� � ���$�
����� ����European Journal of Political Economy, vol. 17, p. 749-777.
[12]'�� ������0��'���!���0��C���>��"�������0�4/???5��,�!��������*�������
World Bank Working Paper No. 2195.
[13]�������0�4/??65�������� ��������#����Quarterly Journal of Economics
110, p. 681-712.
[14]����"����0���������<������0'0�4/??A5�-�� ! �������������.�����
������������������ ��� ������$� ����������*������)������� 38, 379-353.
[15]����"����0�0��������<������0'0��4/??.5�������� ��� ������$%�/���
������������������.����� �����������Journal����*������)������� 65, 520-538.
[16]��<�����'0��������D������80�0�4/???5����������� ��� �����#���������� ��
� ������$����������������������! ������, ����������*������)������� 40;
[17]� !�����<����0� �����!��4+,,65����.��$�������� �� ��/ �����
������ �, Peabody Journal of Education, 80(1), p. 81-92.
[18]� !�����<����0�� �����!��4+,,+5������� �� ��/ ����������� �.,
Retrieved XXmonth dayXX, 2004, from http://admin.corisweb.org/files/
Rumyantseva2002C_Higher_Education1064566056.doc
[19]������������0��E����!���0������� �, Quarterly Journal of Economics 59,
1993, p. 599-617.
[20]�������0�4+,,;5������� ������0�������� 1������� ��,!���������World
Bank Policy Research Working Paper, No. 3824.
[21] ��������0����������0������"���0�������������0�4+,,A5���������� ��� ����
����������� ��*�����# ���&��� ��� !��)�� ������ ����������$� �����������
������ ��'���! ����������+�!������������� ���/ ����������� ���$������
Journal of Economic Computation and Economic Cybernetics Studies and research,
issue 3-4, ASE Publishing House, Bucharest;
[22]���@���E0�4/??-5������� ������������2���%����������� ��������
�������������, IMF.
[23]7�����0�4/??A5��2�$� �������� ���������������. ����������.3�
��� ���� �����" �����Working paper no. 6255, National Bureau of Economic
Research, Cambridge, Massachusetts.
[24]7�����!���0�0�4/??-50��(���������� �����# �������� ���������������
��������%�����! �#� ����������*������)������� 39, 235-274.
[25]*** ������ ��������� ��������������������������������� ( 2004),
Brussels.
[26]Strategia actualizată a Guvernului privind accelerarea reformei în administraŃia
publică 2004-2006, (2004),BucureHti.
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
�
��������������������� ������������ ������������������������������������������������ �����������������������*����������������� �������������������������������� ���������
**���
�
��
��
�
�
�
� ��� ����� ��� �� ���
� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
�����������
�������
�
�
�
� *
(0.095)4.208 ��
*
(0.092)4.209
��� � *
(0.296)4.664 � � *
(0.142)3.921 � � *
(0.254)3.957 � � *
(0.119)4.169 �
��� *0.271− �*
(0.028)0.171− ���
*
(0.027)0.178
�− ��
*0.357− �*
(0.085)0.267− � �� �� *0.329− �
*
(0.084)0.262− �
*0.271− �*
(0.035)0.173− �
��� *0.280− �*
(0.029)0.188− ���
*
(0.028)0.177
�− ��
*0.312− �**
(0.092)0.218− �
*0.425−
�
*
(0.065)0.305− �
� � *0.257− �*
(0.036)0.165− �
2 � � 0.346 ���������
0.342� �
� 0.411 � � ������ � �� ��� � �� ��
+ � � 55.146 ������55.15� � � 11.56 � � ������� � ��!�"� � ������
#$�������
������
� 813 ����������873� � � 116 � � ���� � !�� � ����
#$�������
%� ���������
� !!�������������971� � � � �� � �� � � !�� � ��&�
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
��
����� ������������������� ������������ ��������������������������������������������*����������������� �������������������������������� ���������
**����
*** ��
****��
*****�&�
******!��
��
� ��� ���� ��� �� ���
� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
�����������
�������
�
� *
(0.244)2.369 '
*
(0.237)2.398
�� � *
(0.794)2.332 � � *
(0.380)1.390 � � *
(0.449)2.467 � � *
(0.242)3.048 �
� � *0.336 �*
(0.036)0.169 '
*
(0.035)0.179
��
*0.382 �****
(0.131)0.262 �
*0.380 �***
(0.099)0.230 � � � *0.557 �
*
(0.044)0.196 �
��� *0.305 �*
(0.045)0.209 '
*
(0.044)0.211
�� � *
(0.151)0.396 �
*0.434 �*
(0.126)0.337 � � � � �
��� *0.207− �*
(0.029)0.096− '
*
(0.028)0.102
�− � � *
(0.100)0.215− � � � *0.301− �
*****
(0.048)0.158− �
*0.275− �*
(0.038)0.117− �
�"� *0.235− �**
(0.025)0.061− '
**
(0.025)0.062− � � � � � � � *0.320− �
*
(0.030)0.107− �
(!� *0.214 �*
(0.035)0.099 '
*
(0.034)0.080
�� � � � � *0 .3 4 8 � *
(0.117)0.407 �
*0.204 �****
(0.043)0.083 �
���� *0.129− �**
(0.045)0.110− '
**
(0.044)0.102
�− �
*0.232− �**
(0.139)-0.329 �
*0.190 � � � � � �
���� *0.205 �*
(0.046)0.170 '
*
(0.045)0.152
��
*0.230 � � *0.255 �******
(0.111)0.200 �
***0.189 �***
(0.167)0.368 � 0.193 � *
(0.057)0.158 �
2 � � 0.464 '0.211� � � 0.584 � � ������ � ���� � � ������
+ � � 29.899 '30.684� � � 11.101 � � ���&��� � "�� �� � � ��!��
#$�������
%�����
������
� 768 '808� � � 90 � � �!� � ""� � �""�
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
�
����� ���� � ������ ���������������������
� ��� ���� ��� ��
� �������
���� ����
�����������
) ��*��
�����������
) ��*��
��������
���� �����
�����������
) ��*���
�����������
) ��*���
�������
���� ����
����������� �������
���� ����
�����������
) ����
�������
�
� *
(0.250)3.080 �
*
(0.262)3.621 � � *
(0.382)4.273 �
*
(0.628)4.485 � � *
(0.508)3.805 � � *
(0.544)3.990 �
��� *0.323− � � *
(0.034).166− �
*0.464− �*
(0.071)0.292− � � � � *0.295− �
***
(0.092)0.189− �
� � *0.222− �***
(0.039)0.081− � � � � � � � ****0.258− �
*****
(0.128)0.239− �
+ ���� **0.183− �****
(0.031)-0.053 � � *0.323− �
***
(0.093)-0.266 �
***
(0.104)0.231− � � � � �
��� *0.185− � � *
(0.038)0.095− �
*0.272− � � *****
(0.137)0.294− �
****0.224− �*****
(0.127)0.234− � � �
+����� *0.193− �*
(0.029)0.099− �
*
(0.028)0.097− �
*0.317− � � ***
(0.087)0.146− �
*****0.164− �*****
(0.074)0.135− �
*0.227− � �
(&� *0.232 �*
(0.025)0.095 �
*
(0.024)0.094 �
*0.261 �***
(0.056)0.118 �
***
(0.060)0.123 � � � � �
�"� *0.277 �*
(0.031)0.136 �
*
(0.030)0.113 �
*0.335 � � ****
(0.081)0.156 � � � � �
(!� *0.195− �**
(0.044)- 0.107 �
***
(0.042)0.094− � � � � � � � �
��� *0.257 �*
(0.034)0.142 �
*
(0.034)0.111 �
*0.325 �*****
(0.089)0.162 � � *****0.197 �
*****
(0.088)0.156 �
****0.269 �***
(0.101)0.206 �
���� *0.118 �***
(0.053)0.113 �
*
(0.052)0.103 � � � � � � � �
2 � � 0.438 � 0.460 � � 0.581 � ����!� � ������ � 0.421 �
+ � � 21.210 � 24.721 � � 12.889 � "�"��� � ���&� � 5.245 �
#$�������
������
� 744 � 744 � � 105 � 105 � � 95 � � 76 �
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
���� ��������� ��* *
����������������� �������������������������������� ���������**
����***
��****
��*****
�&�******
!��*����������������� �������������������������������� ��������
**����
***���
****�
*****��
******�
� �� ���
� ����������� ���� �����������) ���� �����������) � �� ����������� ���� �����������) ��,-� �����������) ��,-� �����������) � ,-�
�������
�
� *
(0.521)3.843 �
*
(0.494)4.747 � � *
(0.261)3.262 �
*
(0.262)3.436 �
*
(0.235)3.254 �
��� *0.295− � � *
(0.093)0.189− �
*0.324− �*
(0.044)0.178− �
*
(0.045)0.188− �
*
(0.045)0.198− �
� � ****0.258− �***
(0.127)0.285− �
*
(0.129)0.257− � � � � �
+����� *0.227− �****
(0.074)0.146− �
*
(0.075)0.131− �
*0.142− � � � *****
(0.038)0.069− �
(&� � � � *0.249 �*
(0.031)0.126 � � �
�"� � � � *0.304 �*
(0.039)0.152 �
*
(0.040)0.157 �
*
(0.039)0.162 �
(!� � � � *0.197− �*
(0.052)0.114− �
**
(0.053)0.122− � �
��� ****0.269 �***
(0.100)0.223 � � *0.248 �
*
(0.043)0.106 �
*
(0.043)0.136 �
*
(0.043)0.137 �
2 � � 0.417 � 0.407 � � 0.451 � ���"� ���� �
+ � � 5.115 � 4.826 � � 23.510 � ��� ��� � �"� �
#$�������
������
� 76 � 76 � � 465 � 465 � �&��
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
����������� ������������������ ��������������������������*����������������� �������������������������������� ��������
**����
*** ��
****��
*****�&�
******!�
��
�
�
�
�
� ��� ���� ��� �� ���
� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
�����������
�������
�
� ***
(0.218)0.414 �
�
(0.640)0.602 �
� *
(0.358)0.541− �
� *
(0.455)0.157 �
� *
(0.283)0.555 �
��� *0.697 �*
(0.035)0.703 �
*0.763 �*
(0.094)0.729 �
*0.754 �*
(0.090)0.723 �
*0.593 �*
(0.127)0.697 �
*0.692 �*
(0.044)0.669 �
� � *0.393 �*
(0.034)0.124 � � � *0.412 �
**
(0.081)0.214 � � � *0.412 �
*
(0.044)0.141 �
��� *0.318 �**
(0.041)0.106 �
*0.444 �**
(0.131)0.231 �
� � � � *0.283 �*
(0.050)0.083 �
��� *0.368− �*
(0.027)0.112− �
*0.563− �*
(0.084)0.237− � � � � � *0.360− �
*
(0.034)0.107− �
(!� � � � � *0.415 �***
(0.095)0.216 �
*0.354 �**
(0.141)0.248 � � �
���� *0.289− �**
(0.021)0.056− �
� � � � � � *0.356− �****
(0.028)0.069− �
���� � � � � *0.233− �***
(0.042)0.090− � � � � �
���� *0.238 �****
(0.042)0.108 � � � � � � � *0.279 �
*
(0.054)0.169 �
2 �� 0.736 � � 0.800 � � 0.797 � � 0.614 � � 0.742 �
+ � � 141.28 � � 50.87 � � 36.96 � � 22.44 � � 86.39 �
#$�������
������
� 725 � � 89 � � �!� � "&� � � ��
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
������������ ���������������������������������������������������������������� ����*����������������� �������������������������������� ��������
**����
*** ��
****��
*****�&�
******!�
� ��� ���� ��� �� ���
� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
����������� �������
���� ����
�����������
�������
�
� ***
(0.220)2.093 �
�
(0.587)2.976 �
� *
(0.417)1.098 �
� *
(0.601)2.839 �
� *
(0.262)1.773 �
��� *0.393 �*
(0.032)0.183 �
*0.489 �*
(0.070)0.125 �
*0.412 �**
(0.089)0.155 � � � *0.412 �
*
(0.040)0.207 �
��� *0.364 �*
(0.044)0.278 �
*0.384 �**
(0.121)0.268 �
*0.462 �**
(0.119)0.416 �
*0.351 �**
(0.136)0.345 �
*0.353 �*
(0.052)0.264 �
��� *0.284− �**
(0.031)0.062− �
*0.458− �**
(0.082)0.143− �
� � *0.335− �*
(0.087)0.203− �
� �
(&� *0.209− �*
(0.027)0.086− �
*0.403− �*
(0.076)0.129− �
*0.257− �**
(0.062)0.158− � � � *0.182− �
*
(0.033)0.064− �
("� � � *0.360− �**
(0.082)0.141− �
*0.107 �**
(0.087)0.206 � � � � �
�"� *0.262− �*
(0.023)0.060− �
� � � � � � � �
�!� *0.184 �**
(0.023)0.041 � � � *0.287 �
**
(0.064)0.177 � � � *0.226 �
****
(0.027)0.079 �
��� � � � � � � � � *0.241− �*
(0.039)0.074− �
���� � � � � � � � � *0.303− �*
(0.029)0.074− �
���� � � � � � � � � *0.127 �*
(0.032)0.070 �
���� *0.150 �****
(0.046)0.085 � � � � � � � � �
2 �� 0.509 � � 0.625 � � 0.620 � � 0.428 � � 0.534 �
+ � � 33.87 � � 10.79 � � 10.51� � 8.278 � � 24.45 �
#$�������
������
� 685 � � 89 � � !�� � "&� � � &�
Econometric Models Used for the Corruption Analysis
__________________________________________________________________
������� ������������������������������������������������������������� ������ ������!����������
"�������� #�������������������������� $� ���������������������������
� � � �������� ���������� %�����&�������������������
�������������&�'�������� ������������
������������ 1#− �
(�����������������������������)������������������������������'����������������������������������� �� '������ �� ������������ ����������� � ���� )��������� ��� ���� '��'����*� ��+�������� ����������' ���� �������&*� ������������&� '����� ���������� ����,�����&���� ���� �������������������������������������������������'���������
-� ��
�� .����'�����&�����������������'�������� ������������
������������� 2#− ����
2#′ �
(�� ��� �� ��������� ����� �������������� �����/�������*��''������������� ������������� ���� ����������������������������� �)����'�������������������������� '��&�����������������������������'����������� ���������'�������� ��������������������������
� ��
�� 0��������'����&�����
'����������&��� � 6( )( �
(�� �������� ���� � '���� ��� ���� '��������� �&��� � ��� ���� ������������ ��� ���� ��� '�������� �����������*��������������� ��������� ����������������� ����������'�����������������������������'�������� ����������������������������
1� ��
1� .������� ���������������&��� ��������'���������
�������� 7( )( ��
.������������� ����������� ��� ��������� ��� ������������ ���'�������� �����������������������������������������������������*�����������������)�'�����������������������������&� ����� ���*����� ����� ��������������������������'����&��
�� ��
� !�'����&����������'�������� ������������������������������������
8( )( �
(�� ��� �� ����������������� ��� ��������� ��� ���� ��'����&����'�������� ������������ ��� ������� ���� ���������������������� ��� ���� ��� � ���� ����� �������'����������������'������ ��������� �����������*�����'������������'��������������������������������������*������������������*����������*��������*� '�����������*� ������'���������*� ��������������������������� ���������' ���*� ���������������������'����&��
�2� ��
�� 3���������������'�������� ������������
� '��&���� 3( )# �
(�� ��� ������������������������������� �������������������� ����� '��&��������� ���� ������������'���� � �����&������������� �*��������'�����������������*�����4���*����������������*����� ����������������������������������*������������)���������������������������&�����&�����������������������
�� ��
5� %�����&�������)�
���������� 4( )# �(�������������������������������������,�����&�����������)������������������������������ ������� �� ��'��� ���� ���� ��� � ������ ��'��� ����*� ���� ������� ����� ���� ���� ����� ��� ���������������*� �������� ���� '������� ��� � ������ �� ����� ������������� ����� ��� � ���&� �� �� �����������
�� ��
�� ��������������'����� (�� ��� �� ��������� ����� �������� ���� ������ ��� �����'����� ��� ���� '������ ������� ��� � ���� �� ��
Tudorel Andrei, Stelian Stancu, Monica Nedelcu, Ani Matei
__________________________________________________________________
5( )# � '���'����������'�������� ������������� '��&�����
-� 6���������������'������������� ��������������
������ 6( )# �
(�� ��� �� ��������� ����� ��)��� ����� �������������� ���� ��������� �������� ��� �����'����� ��� ��������� ��������������������' ����������������&������������������' ��������� �������������������������������
5� ��
�2� (���������������������������������������������������������'������
7( )# �
.�������������,�����������������������������������������������������������������������'���� � ����*�������*�������*�'��������������*������������������������4������
�� ��
��� 7������������������������������������������������������'�����
8( )# �
7�� ���������� ��������� ������ �������� ���� ���������&� ��� ���� �������� ����� ����� ��� ����������������������'�����
�� ��
��� 8����������������� '�������������������� ���9'�� ���������������������������
9( )# �
(�� �������������� �����'�������� ����� ���������� '���������'���������������� ����������� �������� '�� ������ ��� ���� ������ �������*� ����� ��� ���� ���������� �/�������� �������� �������������������������'�������������������������� '����������
�� ��
��� %�����&������������������
����� � 10( )# �
(��������������������� �����������������������/���������&���������� �'����������� '���������'�������� �����������������*�����������������&�����'��'��� ���������������''�&���� ������������������ '��������*� ����������� ���� ������&� ��� ������ ��������*� ���� ����������� �������������������������������������������������������������������'������
�� ��
�1� 0������������������������
11 15( )# #− �
.���'�����:��������� 10( ),# ��������� '������&����� 11( ),# ��������������&�; �������������
���� ���������� �����<� 12( ),# � ������ ��� ��������� ;����� ������� �������*� '���� ����� �������
�������*���������������*� �����:�*�0�#<� 13( ),# ����'�����:����������� 14( )# ����������&'��
��� ������������ ��� ������ ���� '������ �������� ���� ���9���� ����������� ;�������� '������
�� �����������*�0�������:�=������*�!����&�!����������������������������������<� 15( ).# �
�� ��